SEASONAL EFFECTS ON VEHICLE MOBILITY: HIGH-LATITUDE … · Page 2 of 8 to characterize seasonal...

Unclassified

2019 NDIA GROUND VEHICLE SYSTEMS ENGINEERING AND TECHNOLOGY

SYMPOSIUM MODELING & SIMULATION, TEST & VALIDATION TECHNICAL SESSION

AUGUST 13-15, 2019 - NOVI, MICHIGAN

SEASONAL EFFECTS ON VEHICLE MOBILITY: HIGH-LATITUDE CASE STUDY

Taylor S. Hodgdon1, Sally A. Shoop PhD1, Susan Frankenstein PhD1, Matthew F.

Bigl1, and Michael W. Parker1

1Engineer Research Development Center - Cold Regions Research and Engineering

Laboratory, Hanover, NH

ABSTRACT Seasonality plays a key role in altering the terrain of many military

operating environments. Since seasonality has such a large impact on the terrain,

it needs to be properly accounted for in vehicle dynamics models. This work

outlines a variety of static and dynamic seasonal terrain conditions and their

impacts on vehicle mobility in an austere region of Europe. Overall the vehicles

performed the best in the dry season condition. The thaw season condition had the

most drastic impact on mobility with all but the heavy tracked vehicle being almost

completely NOGO in the region. Overall, the heavy tracked vehicle had the best

performance in all terrain conditions. These results highlight the importance of

incorporating seasonal impacts on terrain into NRMM or any vehicle dynamics

model. Future work will focus on collecting more data to improve the empirical

relationships between vehicles and seasonal terrain conditions, thereby allowing

for more accurate speed predictions.

Citation: T. Hodgdon, S. Shoop, S. Frankenstein, M. Bigl, M. Parker, “Seasonal Effects on Vehicle Mobility: High-

Latitude Case Study”, In Proceedings of the Ground Vehicle Systems Engineering and Technology Symposium

(GVSETS), NDIA, Novi, MI, Aug. 13-15, 2019.

1. INTRODUCTION Many military operating environments experience

a wide range of seasonal conditions which can have

a significant impact on the ground terrain. This

impact is perhaps the greatest in high-latitude

regions where seasonal climatic changes are more

pronounced. Off-road vehicle mobility is of

particular concern in these regions as vehicles can

be hindered by the degraded ground conditions.

Thus, delivering accurate predictions of vehicle

capabilities in these conditions can provide the

warfighter with crucial information for route

planning analyses. Previous work relating to off

road mobility has focused primarily on static, or

preset, terrain parameters to assess the impacts of

base terrain on off road vehicle mobility. However,

since seasonality has a large impact on the terrain,

it needs to be properly accounted for in vehicle

mobility models. Several studies conducted by

researchers at the Cold Regions Research and

Engineering Laboratory (CRREL) have attempted

DISTRIBUTION A. Approved for public release:

distribution unlimited.

Proceedings of the 2019 Ground Vehicle Systems Engineering and Technology Symposium (GVSETS)

Seasonal Effects on Vehicle Mobility: High-Latitude Case Study, Hodgdon, et al. - Unclassified

of 8

to characterize seasonal impacts on a suite of

vehicles [1-4]. These focused on the generation

and incorporation of new high resolution winter

terrain datasets including: frost depth, thaw depth,

snow depth/density, and river/lake ice thickness.

This particular work outlines a variety of common

static and new dynamic seasonal terrain conditions

and their impacts on vehicle mobility in an austere

region of Scandinavia. A static terrain input dataset

was used to establish a baseline for vehicle mobility

in the region. These static terrain parameters

include: soil class, land usage, and slope. To

highlight the effects of seasonality, three separate

terrain conditions were modeled: dry season soil

conditions, average winter season conditions, and

thaw season conditions. In addition to the static

terrain parameters that were used, each terrain

condition had one or several additional seasonal

parameters added. These parameters include: soil

moisture (for unfrozen conditions), snow

depth/density and frost depth (for winter season

condition), and thaw depth (for thaw season

condition). In addition to the varied terrain

conditions, a suite of vehicles was used for this

analysis to assess the performance of common

military vehicles. The performance of the test

vehicles will be outlined in the Results section of

this report.

2. PREVIOUS WORK As mentioned in the previous section, many

previous analyses focused primarily on assessing

vehicle performance in static terrain conditions.

However, research conducted at CRREL over the

past several decades has focused on developing

new algorithms for seasonal predictions with the

North Atlantic Treaty Organization (NATO)

Reference Mobility Model (NRMM) [5-9].

NRMM is a vehicle mobility platform that

combines several mobility-related technologies

into a single comprehensive model to predict on

and off road mobility capabilities [10]. Winter

mobility algorithms were developed for NRMM

from extensive vehicle testing to characterize the

relationships between a suite of vehicles and varied

seasonal terrain. These algorithms predict the

motion resistance and maximum available traction

for vehicles on winter terrain, including: snow, ice,

and frozen/thawing ground. In addition to field

testing, several studies have been conducted using

vehicle simulators to assess vehicle and driver

performance in winter conditions [8 and 9].

While these studies provided crucial insight into

the relationships between vehicles and the

underlying winter terrain, many of the analyses

utilized uniformly distributed seasonal parameters,

which does not effectively capture the variability of

the terrain [5 and 11]. More recently, studies have

focused on developing new models and methods to

provide geospatially and temporally varying

predictions of these winter season parameters.

These new additions allow for more robust seasonal

terrain inputs to be generated for the NRMM,

thereby providing more accurate mobility

predictions.

3. METHODOLOGY In order to generate robust predictions for vehicle

performance in a variety of cross country

conditions several steps are required (Figure 1).

These steps include:

1) Field Testing to characterize the relationships

between vehicles and the underlying terrain.

2) Analyzing the climatology for an area of

interest to inform selection of the appropriate

seasonal parameters.

3) Conducting geospatial analyses to generate

the input terrain parameters and construct a

representative digital terrain database.

4) Modeling vehicle performance using the input

terrain databases and geospatially distributing

the results.

Steps 1-3 in this methodology will be outlined in

further detail in the following subsections of the

report. The modeling in step 4 was conducted using

NRMM II – CRREL Beta Version 3.0.



of 8

3.1 FIELD TESTING For this analysis, researchers at CRREL have

assessed vehicle performance in winter conditions

at several field sites. These studies included:

testing heavy tracked recovery vehicle capabilities

in Alaska, assessing over snow performance of

light tracked vehicles in Norway and Vermont, and

characterizing wheeled vehicle abilities in deep

snow conditions in Montana. To assess vehicle

performance, a variety of tests are used which

include: motion resistance, drawbar pull, coast

down, and slope testing. The motion resistance test

is used to gauge a vehicles resistance to motion for

a given surface condition. The drawbar pull test

provides an estimate of the possible net tractive

force for the ground conditions. The coast down

test is another setup that is used to test the motion

resistance of a vehicle. Finally, the slope testing

determines how capable a vehicle is on different

grades; both perpendicular and side slop

capabilities are tested. Vehicle performance in

thawing soils was previously characterized by [8,

12-14], but no further field analyses were

conducted as part of this study.

In addition to the vehicle testing, there are also

extensive measurements taken of the underlying

terrain. For measuring the strength of the soil, a

shear vane tester (shear strength), Clegg impact

hammer (bearing capacity), and dynamic cone

penetrometer (bearing capacity) are used. In

addition to soil strength, volumetric soil moisture is

measured using a Field Scout moisture meter. To

characterize the snow properties tests are conducted

to measure common parameters such as: density,

moisture content (Finnish snow fork and Denoth

moisture meter), and depth (Magnaprobe). The

strength of the snow is also measured using the

same instruments as the soil tests, in addition to: the

Rammsonde penetrometer, Yamaha Drop Cone,

CTI snow compaction gauge, and the Snow Micro

penetrometer depending on the amount of

compaction.

Characterize terrain/vehicle interactions

Analyze physical terrain

conditions

Develop vehicle models

Construct terrain database

Predict and validate mobility results

Figure 1: Overview of seasonal cross country mobility

workflow. Including: characterizing terrain/vehicle

interactions, analyzing the physical terrain, developing

vehicle models, constructing terrain databases, and

predicting/validating mobility results



of 8

3.2 CLIMATOLOGY ANALYSIS Climatology analyses play a key role in these

studies as they allow for the proper identification of

representative seasonal terrain conditions. The

analysis for this study used weather data from the

World Bank Climate Change Knowledge Portal

(CCKP) and National Centers for Environmental

Information (NCEI) which include: The Global

Summary of the Day (GSOD), the Global

Historical Climatology Network (GHCN), and the

Integrated Surface Database (ISD). These

databases provide statistical information relating to

temperature, precipitation, and a suite of other

climatological parameters at a fine temporal

resolution over a large period of record (POR). For

this analysis a POR was analyzed from 1961-1990

and another from 1991-2016. This step was crucial

to see if there were significant differences in more

recent years due to globally changing climate.

Generally, both the average air temperature and the

total precipitation are higher in the more recent

POR.

This analysis provided insight into the timing of

key seasonal changes in northern Scandinavia.

These records indicate that the wettest months tend

to be July and August, with the driest month being

May. The freezing season was found to occur

between Mid-October and Early-April with the

thaw season reaching its peak in Mid-April. The

greatest snow depths were observed during the

beginning of March, with the depths increasing as

elevation increased.

3.3 GEOSPATIAL ANALYSIS For each area of interest, geospatial data (such as

satellite imagery, slope, land usage, and soil type)

was gathered to provide a baseline terrain condition

(Figure 2 a-d). Once the baseline was established,

varying seasonal conditions were then

incorporated. One example of this is soil moisture,

which can fluctuate depending on the season, and

is a crucial indicator of soil strength (Figure 2e).

The soil moisture data was generated by Creare

LLC [15]. This dataset provides daily global 30-

meter volumetric moisture predictions, which

allows it to be used for any area of interest.

Winter plays a significant role in vehicle mobility,

so new methods had to be developed for generating

terrain datasets in winter conditions. These datasets

typically include frost depth, snow depth, ice

thickness and snow density (Figure 3). These

geospatially distributed seasonal datasets are

generated using a variety of models developed at

CRREL. SnowModel [16] is used to predict the

snow depth and density for a given area and time

Figure 2: Examples of the parameters that are used to

construct a terrain database. A: Satellite image of the test

area in northern Scandinavia, B-D: Static terrain

parameters, and E: soil moisture example of dynamic

seasonal inputs.

A

B C

D E



of 8

period. This model was selected as it is a snow

evolution modeling platform that allows for

spatially distributed results with high spatial (1-

200-meter cell size) and temporal resolution (10

min-1 day). Required inputs to SnowModel

include: topography and land cover (vegetation

type) as well as time dependent climatic fields such

as precipitation, wind speed, air temperature, and

relative humidity. The climatic inputs are typically

obtained from meteorological stations or an

atmospheric model.

In addition to snow characterization, the strength

of the underlying soil is also crucial for correctly

preceding vehicle capabilities. The depth of frost

can change the competency of soil considerably.

The MODBERG (Modified-Berggren) model [17]

is used to calculate estimated maximum frost depth.

This model takes into account the surficial soil type

and estimated gravimetric soil moisture, as well as

the average air freezing degree days and snow

depth for a particular region. These parameters are

used with the Modified-Berggren equations to

calculate the maximum frost depth. This model

does not inherently produce geospatially

distributed results. However, the results were able

to be mapped following a modified geospatial

distribution process. Before MODBERG is run, the

input parameters are geospatially mapped in the

region of interest. These parameters are spatially

joined into a single raster dataset, with unique

numeric identifiers created for each combination of

input parameters. This process is similar to the

generation of Numbered Terrain Units (NTU’s) for

which NRMM is required to run. The attribute

table of this spatially joined raster dataset is

exported, and its contents are used as the input

parameters to MODBERG. A maximum estimated

frost depth is calculated for each combination of

input parameters and the results are mapped back

by matching the unique identifiers to the input

raster dataset. Another equally important factor

affecting soil competency is the depth of thawing

ground. The FROSTB model [18] was used to

estimate the average thaw depth for a particular

point in time. This model is not setup in a manner

to accept geospatial inputs thus, to represent the

thaw depth, a value of 5.5 inches (calculated using

FROSTB) was uniformly distributed across the

area of interest. This value was selected as a proxy

to represent the average depth of thawing soil in the

test region. The majority of the static and dynamic

datasets in this study had a spatial resolution of 30

meters, with the exception of the snow

depth/density which had a 400 meter resolution.

Once all of the seasonal inputs were generated

they were spatially combined to the static terrain

database. The seasonal condition being modeled

dictated which data layers would be used to

generate the input file for NRMM. The output

attribute table generated from the spatial combine

was used as the input into NRMM.

B

C E

Figure 3: Examples of seasonal parameters used. A:

Satellite image of the northern Scandinavia test area in

winter, B: modeled maximum frost depth, C and D:

modeled snow depth and density.

C

D

A

C



of 8

4. RESULTS AND ANALYSIS The results of this analysis indicate that there is a

drastic impact on vehicle speed depending on the

season (Figure 4). Overall the vehicles performed

the best in the dry season condition, which is to be

expected since the terrain is generally more

competent. In the dry season, there is lower

volumetric soil moisture, which increases the

strength of the soil. The exception to this is with

sand-dominated soils, where the strength increases

with increasing moisture content. The thaw season

condition had the most drastic impact on mobility,

with all but the heavy tracked vehicle being almost

completely NOGO in the region. The winter season

showed the slightly hindered vehicle performance

due to the snow on the terrain. However, all of the

vehicles were able to traverse peat soils due to the

fact that most of the ground under the snow was

frozen.

Overall, the heavy tracked vehicle had the best

performance in all terrain conditions. In the dry

season condition, the heavy tracked vehicle was

hampered by peat soils. However, the greatest

hindrance to speed for the heavy tracked vehicle

was the slope of the terrain, the effect of which was

exacerbated in the winter season. This decrease in

slope mobility is due to there being less traction

between the vehicle and the underlying terrain. The

performance of the light tracked vehicle seemed to

be hampered mostly by the amount of vegetation

due to its low push-over force compared to the

other vehicles. However, this vehicle was the only

one capable of traversing peat soils in the dry

season condition. Similar to the heavy tracked

vehicle, the light tracked vehicle was mostly

hampered by the slope of the terrain, especially in

winter conditions. The light and heavy wheeled

vehicles had similar performance to one another in

all conditions. While performance of both vehicles

was impacted by the slope of the terrain, the light

wheeled vehicle performed better on slopes in the

dry season than the heavy wheeled vehicle.

However, the opposite was true for the winter

season condition. This is likely due to the larger

snow depths at higher elevations. The heavy

wheeled vehicle has a higher ground clearance and

was likely not experiencing as much drag as the

light wheeled vehicle.

5. CONCLUSIONS These results highlight the importance of

incorporating seasonal impacts on terrain into

NRMM and Next Generation NRMM. All of the

test vehicles experienced impacts to performance in

both the winter and thaw season conditions

compared to the dry season. Overall, the greatest

hindrance to speed for all four test vehicles was the

5.5 inches of thawing soil. In winter season

conditions, the ground beneath the snow was frozen

enough to allow vehicles to safely traverse across

the areas of peat. Future vehicle technology needs

to be adapted to be suited for the terrain of austere

northern regions. These updates include: reviving

deep and over snow vehicle capabilities, addressing

current science and technology gaps, applying new

modeling and simulation techniques, and

developing autonomous vehicle capabilities. In

addition to updating vehicle technology, future

modeling and geospatial work must focus on:

quantifying and incorporating uncertainty into

speed predictions, adding capabilities for peat and

highly organic soils, addressing vegetation effects

on terrain strength, and validating updated

algorithms that quantify the effects of seasonally

changing terrain conditions on vehicle mobility.



of 8

Figure 4: Seasonal speed maps for the dry season, winter season, and thaw season, for all four test vehicles.

50 – Max Speed

30 – 50

15 – 30

5 – 15 0 – 5

Water

Speed Range (kph)

Heavy T

racke

d V

ehic

le

Lig

ht

Tra

cked V

eh

icle

H

eavy W

he

ele

d V

ehic

le

Lig

ht

Wh

ee

led

Vehic

le



of 8

6. REFERENCES

[1] S. Shoop, P. Richmond, and J. Lacombe

“Overview of cold regions mobility modeling at

CRREL”, Journal of Terramechanics 43.1, pages

1-26, 2006.

[2] R. Affleck, R. Melloh, and S. Shoop, “Cross-

country mobility on various snow conditions for

validation of a virtual terrain”, 8th European

conference of international society of terrain

vehicle systems. Umea, Sweden. 2000

[3] G. Blaisdell, P. Richmond, S. Shoop, C. Green,

and R. Alger. “Wheels and tracks in snow:

validation study of the CRREL shallow snow

mobility model”. CRREL Report 90-9. US Army

Cold Regions Research and Engineering

Laboratory, Hanover, NH, 1990

[4] S. Shoop. “Vehicle bearing capacity of frozen

ground over soft substrate”. Canadian Geotech

Journal 32.3, pages 552-556, 1995.

[5] R. Affleck and S. Shoop, “A spatially based

mobility performance model”, Journal of

Terramechanics 46.4, pages 203-210, 2000.

[6] R. Melloh, P. Richmond, S. Shoop, R. Affleck,

and B. Coutermarsh, “Continuous mapping of

distributed snow depth for mobility models using

shaped solutions.” Cold Regions Science and

Technology 52.2, pages 155-165, 2008.

[7] R. Melloh, S. Shoop, B. Coutermarsh, R. Affleck,

and P. Richmond, “Distributed snow properties for

mobility models" Cold Regions Science and

Technology, International Polar Year 2007

Special Issue.

[8] P. Richmond, S. Shoop, and G. Blaisdell. “Cold

regions mobility models.” No. CRREL-95-1. Cold

Regions Research and Engineering Laboratory

Hanover NH, 1995.

[9] R. Ahlvin and S. Shoop. “Methodology for

predicting for winter conditions in the NATO

reference mobility model”. In: 5th North

American conference of the ISTVS. Saskatoon,

SK, Canada, 320-34, 1995.

[10] R. Ahlvin and P. Haley. “NATO reference

mobility model, Edition II, NRMM II user’s

guide”, USAEWES, Geotechnical Laboratory,

Technical Report GL-92-19, 1992.

[11] S. Shoop, R. Affleck, C. Collins, G. Larsen, L.

Barna, and P. Sullivan. “Maneuver analysis

methodology to predict vehicle impacts on

training lands.” Journal of Terramechanics 42, no.

3-4. p 281-303, 2005.

[12] Shoop, S. A. “Vehicle mobility on thawing soils:

Interim report on CRREL's test program” (No.

CRREL-SR-89-31). Cold Regions Research and

Engineering Lab Hanover NH, 1990.

[13] Shoop, S. A. “Effect of soil thawing on off-road

vehicle traction. In Proceedings, 6th International

Permafrost Conference, July 1993, Beijing China.,

vol.1 p 559-563, 1993c.

[14] Shoop, S. A. “An experimental method for

vehicle mobility testing on freezing/thawing soil.

In Proceedings of the First International

Conference on Winter Vehicle Mobility, June

1991, Santa Barbara, California (G.L. Blaisdell,

Ed.). USA Cold Regions Research and

Engineering Laboratory, Special Report 93-17,

1993.

[15] Creare LLC. 2018. Global Volumetric Soil

Moisture [geotiff format]. Retrieved from

https://mobility.crearecomputing.com

[16] G. Liston and K. Elder, “A distributed snow-

evolution modeling system (SnowModel)”,

Journal of Hydrometeorology 7.6, pages 1259-

1276, 2006.

[17] G. Aitken and R. Berg, “Digital solution of

modified Berggren equation to calculate depths of

freeze or thaw in multilayered systems”, No.

CRREL-SR-122, Cold Regions Research and

Engineering Laboratory, 1968.

[18] G. Guymon, R. Berg, and T. Hromadka.

“Mathematical model of frost heave and thaw

settlement in pavements”, No. CRREL-93-2, Cold

Regions Research and Engineering Laboratory,

Hanover NH, 1993.

SEASONAL EFFECTS ON VEHICLE MOBILITY: HIGH-LATITUDE … · Page 2 of 8 to characterize seasonal...

Documents

Transcript of SEASONAL EFFECTS ON VEHICLE MOBILITY: HIGH-LATITUDE … · Page 2 of 8 to characterize seasonal...